Comparative Analysis of Eighteen IOL Power Calculation Formulas Using a Modified Formula Performance Index Across Diverse Biometric Parameters.

Abstract

Purpose: To evaluate the predictive accuracy of 18 intraocular lens (IOL) power calculation formulas across diverse biometric parameters using a modified formula performance index (FPI) and prediction error variability metrics.

Design: Prospective, comparative, accuracy and reliability analysis of IOL power calculations.

Methods: This study included 213 cataract patients who underwent phacoemulsification with SN60WF IOL implantation. Preoperative biometry was obtained using the IOLMaster 700. Eighteen IOL power calculation formulas, including Barrett TK, Barrett Universal II, Castrop, Cooke K6, EVO 2.0, Haigis, Hoffer Q, Hoffer QST, Holladay 1, Holladay 2, Kane, Karmona, Ladas Super Formula, Nallasamy, PEARL-DGS, RBF 3.0, SRK/T, and T2, were evaluated. Predictive accuracy was assessed using FPI and standard deviation (SD) for overall performance, while root mean square error (RMSE) and subgroup-specific FPI (FPI_sub) were used for biometric subgroup analysis. Additional metrics included mean absolute error (MAE), median absolute error (MedAE), and the percentage of eyes within ±0.25 D, ±0.50 D, ±0.75 D, and ±1.00 D.

Results: Cooke K6 achieved the highest FPI (0.724) and lowest SD (0.394), followed by PEARL-DGS (FPI=0.714, SD=0.399) and Barrett TK (FPI=0.710, SD=0.411). Barrett TK had the highest percentage of eyes within ±0.25 D (54.5%). Subgroup analysis revealed that PEARL-DGS exhibited the highest accuracy in short eyes (AL≤22 mm, FPI_sub=0.838, RMSE=0.271), while Haigis performed best in long eyes (AL≥26 mm, FPI_sub=0.826, RMSE=0.336) and steep corneas (K>46 D, FPI_sub=0.765, RMSE=0.350). Barrett TK ranked highest in medium-long eyes (24.5-26 mm, FPI_sub=0.700, RMSE=0.400) and flatter corneas (K<42 D, FPI_sub=0.639, RMSE=0.382). Cooke K6 was most accurate in shallow anterior chambers (ACD≤3.0 mm, FPI_sub=0.737, RMSE=0.383). Traditional formulas, including Hoffer Q, Holladay 1, and SRK/T, exhibited lower accuracy and higher prediction error variability.

Conclusions: Modern formulas, particularly Cooke K6, PEARL-DGS, and Barrett TK, demonstrated superior accuracy with lower prediction error variability than traditional formulas. Subgroup analysis highlighted the importance of biometric-driven formula selection to optimize refractive outcomes, providing a clinically reference for personalized IOL power selection.